Want to take part in these discussions? Sign in if you have an account, or apply for one below
Vanilla 1.1.10 is a product of Lussumo. More Information: Documentation, Community Support.
I am periodically asked what has become of Joyal’s Cat Lab, which seems to be dormant or moribund. Has anyone asked him about this recently?
There would need to be a certain amount of clean-up, but we might need his permission…
Oh, phooey, if I’d thought of it a week ago I could have asked him about it at the VV memorial conference.
Just remind me what the goal would be here? I think it is clear that the project is no longer active. Is it that people would like to begin editing it? Or that we should put up some message that the project is no longer active?
Thanks very much for joining the discussion! By “goal”, I meant just what we would be asking Joyal about.
Personally I don’t think I’ve seen anything at the CatLab that distinguishes the content from what would be suitable for the main nLab. But no problem if others feel differently.
I might be slightly mistaken, but I think Joyal had wanted to put his imprimatur on his CatLab, and thus wanted personal oversight over any substantial edits (such as exposition). I’d bet money he wouldn’t mind at all if someone fixed things like commutative diagrams, but I think we should still ask whether he would mind.
It’s been a long time since my last contact with him, so I’d prefer it not be me to reach him about this, although I will if no one else volunteers.
Okay, I wrote André about this.
And now (as Richard will know when he reads this), André has graciously written back, giving permission to perform fixes.
Password sent to Todd now.
I’ve also switched on nForum announcements for the CatLab, in case Todd or anybody else wishes to make use of these.
To me, I must say (this is not meant as a criticism) that it has never struck me as especially original. I think some of the material on distributors is probably the only thing I have seen that might not be considered standard. But as you say, it does have a certain organisational structure and perspective that comes from being a single person’s cohesive work. I am completely happy for it to exist :-)
Zoran, I didn’t check out that link, but hopefully this Fosco is the same as Fosco Loregian who is a category theorist and friend of the nLab. (Fosco, could you explain a little more about your link?)
Oh, I see: Fosco was linking to TeX-based software for commutative diagrams. This could be useful for nLabbers.
But has nothing to do with Joyal, and i see just smple code out of context, no Readme file, even no name of which language the code is in. Just link thrown into discussion.
A link to a page that makes diagrams in a discussion which includes laments that diagrams no longer work on the CatLab. :-)
I know David, but the link isunusableand this is not only about Catlab. Codecogs is broken in mainkab in my lab and so on. Link to 10 lines if diagram code in unknown language looks like google generated association, rather than a conversational replica. What is the language?
The source code can be found here. It looks like a library for creating commutative diagrams in TikZ, i.e. it is a kind of layer on top of plain TikZ. It is written in LaTeX as far as I see, i.e. it just introduces some convenient syntax.
I am planning to add TikZ support to the nLab as soon as time allows. I have also mentioned before the possibility of having some implementation-agnostic syntax for commutative diagrams; kodi is something along these lines, it seems, although more closely tied to TikZ than I was envisioning.
Implementing the TikZ is a difficult project. If we have it then anyway many paths will be open for further development. The diagram support is of course our main need.
I would like to (gradually, regarding little available free time) like to replace codecogs formulas in catlab with diagrams which now work, could I get an access to editing for this purpose ?
I second this, fwiw
This was already discussed here: https://nforum.ncatlab.org/discussion/1304/joyals-cat-lab/
And there is no need to manually replace diagrams, it can be easily done automatically by a script.
However, I need to see the source code in order to write a correct script, and currently the CatLab is protected from viewing (or editing) the source code.
I have replied on the other thread now. Some time back it was raised whether the CatLab could be opened up, but it was decided against it. Todd has been the de facto person in charge, and I would need approval from him if we were to open it up; if I recall correctly (but I may well not be!), it might be that André Joyal has not given explicit approval for people making non-trivial edits. We have better spam control and awareness now, so I would not be opposed to opening it up from that point of view.
Let me get back to you…
I think all what we need is to perform a one-time edit using a script.
No need to open the CatLab for general editing.
I am now looking into the source code to see how to fix it.
I have managed to produce a complete list of pages that still use codecogs in their latest revision using the following SQL script:
create table latest as (select revisions.* from revisions inner join (select page_id, max(id) as maxid from revisions group by page_id) groupedid on revisions.page_id = groupedid.page_id and revisions.id = groupedid.maxid);
create table codecogs as (select * from latest where content like '%codecogs%');
select page_id, pages.web_id, webs.name, pages.name from codecogs join pages on codecogs.page_id = pages.id join webs on codecogs.web_id = webs.id;
Here is the result:
+---------+--------+-------------------+----------------------------------------------------+
| page_id | web_id | name | name |
+---------+--------+-------------------+----------------------------------------------------+
| 1058 | 1 | nLab | torsor with structure category |
| 2267 | 3 | Schreiber | quantization of differential nonabelian cohomology |
| 3096 | 11 | Zoran Skoda | my links |
| 3229 | 23 | Joyal's CatLab | Categories |
| 3244 | 23 | Joyal's CatLab | Natural transformations |
| 3284 | 23 | Joyal's CatLab | Introduction |
| 3338 | 23 | Joyal's CatLab | Functors of two variables |
| 3425 | 23 | Joyal's CatLab | Factorisation systems |
| 3453 | 23 | Joyal's CatLab | Model categories |
| 3472 | 23 | Joyal's CatLab | Weak factorisation systems |
| 3475 | 23 | Joyal's CatLab | Cartesian squares |
| 3486 | 22 | Domenico Fiorenza | Flat infinity-connections |
| 3487 | 22 | Domenico Fiorenza | Functorial path integrals |
| 3525 | 23 | Joyal's CatLab | Model structures on Cat |
| 3529 | 23 | Joyal's CatLab | Grothendieck fibrations |
| 3531 | 23 | Joyal's CatLab | Distributors and barrels |
| 3613 | 23 | Joyal's CatLab | Spans and cospans |
| 3678 | 23 | Joyal's CatLab | Kan extensions |
| 5684 | 23 | Joyal's CatLab | Adjoint Functors and Monads |
| 9044 | 11 | Zoran Skoda | test page |
+---------+--------+-------------------+----------------------------------------------------+
I fixed torsor with structure category to see if xymatrix works correctly, and it does. Apart from Joyal’s CatLab, the remaining pages are on private webs and presumably can be fixed by their owners.
xymatrix doesn’t work when I view version 8 directly, but it does if I just look at the page normally.
The ’new’ renderer is still not activated on ’historical’ versions of pages, which includes viewing the current page by explicit reference to its version.
So I fixed Weak factorisation systems (joyalscatlab), but then strange new errors (“Invalid LaTeX block”) emerged that I am unable to fix.
What is even more mysterious is that when I attempt to locate the exact reason for “Invalid LaTeX block” using bisection, the problem always disappears.
Specifically, in the code below, if I remove everything before Definition labeled “Sigmafibrant”, then the code renders correctly. And if I remove everything starting (and including) this Definition, the code renders correctly.
But if I concatenate them together (with a few blank lines in between), then the parser produces a mysterious “Invalid LaTeX block” error.
Here is the source: https://pastebin.com/2bZDJXea
And here is the part that I was able to get rendered: https://ncatlab.org/nlab/revision/Sandbox/2076
When the Invalid Latex Block error splurges a huge amount of content that it thinks is suspicious, this is almost certainly due to an incorrectly opened or closed latex block somewhere. It is not always easy to see where this has occurred, but I’d scroll back in the rendered page from where the suspicious content begins to see if any incorrect rendering can be spotted. And indeed here we find an incorrectly formed latex block just before where the phrase ’transfinite op-compositions’ occurs at the end of the proof of Lemma 4.2, and we can see an incorrectly rendered list after that, for example.
I see, so this error was already present in Joyal’s original source. How come nLab was able to display it before that?
Yes, correct. When making a page edit, the renderer which I wrote in 2018 or so is used, which is slightly stricter than the old one. If a page has never been successfully rendered by the 2018 renderer, the old Instiki renderer is used to render on the fly. We can see the rendering errors consequent to the latex block error (the incorrectly rendered list for example) on the current Cat Lab page; the old Instiki renderer still allowed page edits to go through with errors in some circumstances like this, whilst the 2018 renderer does not.
I see, this makes sense now. Revision 2079 is almost what I intended as a replacement for https://ncatlab.org/joyalscatlab/published/Weak+factorisation+systems, but there is a new problem (not present in the original version): the numbering of sections is off by 1, e.g., “Main definitions” should have number 1, but is numbered 2 instead. Do you know why it is this way?
It is due I think to the fact that there is a section header in the context menu, which is becoming 1. No time just now to look into it more, but I can see that the context menu looks pretty awful on the published Cat Lab page too where the old Instiki renderer has been used, so probably something is amiss with it.
Okay, I fixed it by using the newer format for context menus.
I believe that https://ncatlab.org/nlab/revision/Sandbox/2081 can now be transplanted to https://ncatlab.org/joyalscatlab/published/Weak+factorisation+systems
Ok great, I’ll do that when next at my computer (have been using my phone for investigating the last few things in this thread).
https://ncatlab.org/nlab/revision/Sandbox/2083 can now be transplanted to https://ncatlab.org/joyalscatlab/published/Model+categories
I had to (re)submit it 6 or 7 times before the server could accept it without timing out—is such behavior normal?
I will now collect in this post links to all fixed articles. I included the page_id and web_id if you need it. The left column is the revision number in Sandbox.
Okay, I now have all articles ready for upload, but I cannot submit them to Sandbox because the nLab interface merges all recent revisions, so the new article overwrites the previous one. What are the alternatives?
2081 | 3472 | 23 | Joyal's CatLab | Weak factorisation systems | 2083 | 3453 | 23 | Joyal's CatLab | Model categories | 2085 | 3229 | 23 | Joyal's CatLab | Categories | To be processed: 3244 | 23 | Joyal's CatLab | Natural transformations | 3284 | 23 | Joyal's CatLab | Introduction | 3338 | 23 | Joyal's CatLab | Functors of two variables | 3425 | 23 | Joyal's CatLab | Factorisation systems | 3475 | 23 | Joyal's CatLab | Cartesian squares | 3525 | 23 | Joyal's CatLab | Model structures on Cat | 3529 | 23 | Joyal's CatLab | Grothendieck fibrations | 3531 | 23 | Joyal's CatLab | Distributors and barrels | 3613 | 23 | Joyal's CatLab | Spans and cospans | 3678 | 23 | Joyal's CatLab | Kan extensions | 5684 | 23 | Joyal's CatLab | Adjoint Functors and Monads |
Re #40: the nLab server itself probably does accept it, it is a layer in front of the nLab (CloudFlare) which times out. Best just to wait a while after you receive the error and refresh the Sandbox, or whichever page you’re editing, to see if the edit takes effect. These pages are quite long with many diagrams, which slows things down, but hopefully this timeout issue should go away with the ongoing rewrite of the nLab server.
Re #41: Great! You could try submitting to the Sandbox with a different author name each time, like ’Dmitri Pavlov - 3229’, or whatever.
Okay, something is terribly wrong with nLab’s software.
I submitted article 3229 under the user name “Article 3229”. It timed out, but I was able to see the correctly rendered article 3229 as revision 2085 and it was correctly indicated in the history under the user name “Article 3229”.
I then submitted article 3244 under the user name “3244 Article”. It timed out, but I was able to see the correctly rendered article 3244 but it overwrote the previous version completely, since it was written under the same revision number, 2085, and the user name was “Article 3229”!
I actually verified this twice to make sure there is no mistake. So submitting two articles to Sandbox under completely different user names results in the older version being silently overwritten.
Perhaps it’s not the user name but the IP address that is responsible for such overwritings?
I’ll take a look as soon as I get a chance. This doesn’t happen in general/under normal circumstances, so there must be something particular to your case, yes.
I haven’t had time yet to look into the bug unfortunately, but I have now moved Revision 2081 of the Sandbox to Weak factorisation systems. A little code change was needed to use the ’new’ renderer on published pages, which is necessary due to the use of its syntax features (e.g. the use of xymatrix).
Please check that everything looks OK. The context bar could still do with a little editing, it seems, but it is better than before.
It looks like there is a tiny bug with the link colour in the table of contents, and also a bug in that page links are being made to /show rather than /published. I’ll fix those as soon as I get a chance. The main thing for now is to check that I’ve moved the correct revision, and that it looks as expected!
Ah, actually the following is not a bug.
also a bug in that page links are being made to /show rather than /published.
When created using nLab page syntax, the links are correctly being made to /published. It is links to theorems using HTML references which are broken in the above way, and this is because they have been hard-coded.
We could of course fix them manually, but it seems to me that a better fix might be to always redirect /show to /published for a published web; this basically means that published means published for the author of the web and anybody else with the web password, whereas currently they can over-ride and get a page which they can edit, which shows the revision history, etc. If anyone has any objections to/sees any problems with this, let me know, otherwise I’ll go ahead and implement it. An alternative would be to change /show to /published only where needed in links in the renderer.
Thanks, looks good. Redirecting /show to /published is fine with me.
I also have the source files for the remaining pages for CatLab (the ones I could not upload), with fixes for all uses of xymatrix.
I have now made a code change so that /show is treated in the same way as /published for a published web, and links on the CatLab should work now.
Regarding the following in #45
It looks like there is a tiny bug with the link colour in the table of contents
this was not actually a bug either, it was the same with the old renderer; there was simply no CSS to override link colour in the table of contents. I have now fixed it by changing the CSS config for the CatLab. As a side-effect of this, the context bar now displays better on all pages.
I have still not had a chance to look into the page saving bug; it is probably best for me to fix this and for you to continue the pages to the Sandbox once I’ve done that, Dmitri, since this allows you to check that the syntax is valid.
In passing, I see various things in the CatLab that can be cleaned up: there are some pages not by Joyal that can be deleted, some query boxes which can be removed, etc. I will try to do this clean up if I get a chance.
I also see that the table of contents is not fully working due to the use of un-numbered theorem environments; again, I will fix that when I get a chance.
I have now fixed the theorem environments/table of contents at Weak factorisation systems, as well as tidied up the context bars. The page should now be in a good, stable state. If anybody notices anything further that should be cleaned up, or which does not work, let me know!
There are several pages on the CatLab created long after Joyal ceased working on the project by another person ’jstalfos’, such as the following one and the pages linked from it.
https://ncatlab.org/joyalscatlab/published/General+topology
I propose to simply delete this page and all other ones by this author. There does not seem to be anything in the content which is not already on the nLab. Please let me know if there are any objections. I could archive the source of the current versions of the pages-to-be-deleted here.
jstalfos must be this user: https://mathoverflow.net/users/26336/john-stalfos
1 to 51 of 51